| encoded_sequence_a = tokenizer(sequence_a)["input_ids"] | |
| encoded_sequence_b = tokenizer(sequence_b)["input_ids"] | |
| The encoded versions have different lengths: | |
| thon | |
| len(encoded_sequence_a), len(encoded_sequence_b) | |
| (8, 19) | |
| Therefore, we can't put them together in the same tensor as-is. |